Top Tweets for #voxforge

_Gooofy_ @_Gooofy_

over 7 years ago

Finally new English #zamia-speech #kaldiasr models are available for download here: https://t.co/hCCZi8I9PP - trained on 1200+ hours of recordings, 150k words, 7.55% WER. #CommonVoice #voxforge #speechrecognition #ArtificialIntelligence

_Gooofy_'s tweet photo. Finally new English #zamia-speech #kaldiasr models are available for download here: https://t.co/hCCZi8I9PP - trained on 1200+ hours of recordings, 150k words, 7.55% WER. #CommonVoice #voxforge #speechrecognition #ArtificialIntelligence https://t.co/bXJVn9ThnV

1

8

3

0

0

_Gooofy_ @_Gooofy_

over 8 years ago

Our latest English #kaldiasr models come with improved noise resistance and tokenizer fixes resulting in slightly better WERs (6.79% for the big model, 7.78% for the #RaspberryPi one) https://t.co/51SbRAx1oG #speechrecognition #voxforge #AI

0

2

0

0

0

_Gooofy_ @_Gooofy_

over 8 years ago

status: so it _is_ possible to train @mozilla's DeepSpeech on non-AVX CPUs and a single GTX 980 (4GB) GPU :) #voxforge #asr #speech #ai #DeepLearning

_Gooofy_'s tweet photo. status: so it _is_ possible to train @mozilla's DeepSpeech on non-AVX CPUs and a single GTX 980 (4GB) GPU :) #voxforge #asr #speech #ai #DeepLearning https://t.co/2nzjOuFpaq

0

1

0

0

0

_Gooofy_ @_Gooofy_

almost 9 years ago

latest 20171217 release of the german #voxforge #kaldiasr #cmusphinx speech recognition models is out: https://t.co/pdAYyYodVa 1.12% WER for kaldi, 11.3% WER for sphinx #ai #asr

1

3

0

0

0

_Gooofy_ @_Gooofy_

almost 9 years ago

#kaldiasr 5.2 tdnn-chain model trained on german #voxforge data achieves 1.18% WER (!!) and is nearing realtime performance on a #RaspberryPi 3 ! download model here: https://t.co/pdAYyYodVa

_Gooofy_'s tweet photo. #kaldiasr 5.2 tdnn-chain model trained on german #voxforge data achieves 1.18% WER (!!) and is nearing realtime performance on a #RaspberryPi 3 ! download model here: https://t.co/pdAYyYodVa https://t.co/KDpPacL4h8

0

3

1

0

0

_Gooofy_ @_Gooofy_

almost 10 years ago

Kaldi's nnet3 LSTM model is actually still a bit better than the regular nnet3 one: 1.33% vs 1.52% WER! #voxforge #german #kaldiasr

0

1

0

0

0

_Gooofy_ @_Gooofy_

almost 10 years ago

german #voxforge model stats: 30686 lexicon entries. total duration of all good submissions: 214:57:22 (up from 70 hours in last release)

0

1

0

0

0

_Gooofy_ @_Gooofy_

about 10 years ago

german #voxforge model is up to 80 hours, 29k words now: https://t.co/lNsdgmCkXv

0

1

0

0

0

Ahmet Emre Aladağ

over 10 years ago

#Voxforge #SpeechRecognition sistemi için sesli corpus lazımmış. #Boğaziçi #GETEM'dekiler iş görür mü acep? https://t.co/2cw95t57wy

0

1

0

0

0

Jan Vlug 🌱 @janvlug

almost 11 years ago

@AstaMcCarthy @Elfons0 Also Dutch: http://t.co/2NWGQ80lsT. Note that I do not know how active #VoxForge is. I've no hands-on experience.

0

1

0

0

0

Jan Vlug 🌱 @janvlug

almost 11 years ago

Consider contributing to #Linux speech recognition by providing readings to #VoxForge: http://t.co/RpjdfzlrkE https://t.co/XQ8edmUdMw

Elfons @Elfons0

almost 11 years ago

Love Linux but there is sadly no useable alternative for automatic speech recognition https://t.co/5QRdxzaHpe

0

0

0

0

0

1

0

0

0

0

_Gooofy_ @_Gooofy_

over 11 years ago

hey - so cool: looks like my german #voxforge model was mentioned in the #cmusphinx blog about a year ago: http://t.co/D7qSGWWSvL

0

1

0

0

0

ΕΕΛΛΑΚ @eellak

about 12 years ago

Το #VoxForge ιδρύθηκε για να συλλέξει μεταγραφή ομιλίας (#transcription) για χρήση από εργαλεία ... http://t.co/RF5t46ZyTE #opensource

0

4

2

0

0

Last Seen Hashtags on Sotwe

Trends for you

Most Popular Users